Abstract: The phenomena that parking is difficult in large and medium sized cities makes people’s desire for parking lot information become stronger and stronger, and the level of detail and accuracy of parking lot information in electronic maps directly impact the service quality of maps. In view of the problems of current surveying methods of parking lots, this paper proposes an approach to automatically extract the locations of parking lots from floating car data. In this paper, the DBSCAN algorithm is used to detect point clusters located within parking areas, and a special spatial grid index is established to decrease the time complexity of the clustering algorithm. Experiment results validate the proposed method.